Size and compare plant alternatives

Use Sizing to search component sizes under a stated objective and limits. The search evaluates modeled designs. It does not prove that an unbounded physical or commercial optimum exists.

Run and inspect a baseline first. Keep the demand, resource and cost basis clear so a better objective value represents a meaningful comparison.

Prepare the search

Open the plant Design and choose the components that can change.
Set each relevant size to automatic sizing, a fixed value or a search range. Check the lower limit, upper limit and step where offered.
Review equipment costs, lifetime assumptions, dispatch and feasibility limits.
Save the configuration and open Sizing.
Choose the search algorithm: grid search, genetic algorithm, particle swarm or NSGA-II. Select it for the search size and comparison you need; a heuristic is not a guarantee of the global optimum.
Choose the objective. The default mean_lcoe path evaluates one year. Percentile and worst-tail objectives require the stated multi-year weather window and cached years.
Review the simulation budget and seed before starting. Retain the seed when you need to reproduce a stochastic search.

The Design shortcut Search for a cheaper size saves a changed version, searches its sizes, and adopts and simulates the winning result through that workflow. Use the dedicated Sizing controls when you need to inspect search choices explicitly.

Read the answer

Read feasibility before rank. If no candidate satisfies the limits, the completed answer can be No feasible solution. Inspect the named binding constraint and closest miss. Do not weaken a reliability requirement solely to produce a green result.

Use the solution table and Pareto chart to compare trade-offs. A Pareto alternative has no other displayed candidate that is at least as good in every selected objective and strictly better in at least one. It is not an approval.

Compare up to three candidates in the comparison view. Read component sizes, reliability, resource use and costs together. Keep the assumptions alongside the headline values.

Save the chosen alternative

Use the promotion action for the selected solution to retain it as a design version. Check the resulting version and simulate it to review the operating year.

For a search bound to network elements, the result also carries a sizing proposal. Apply that proposal before another model change makes it stale. Follow Plant to network; do not copy old proposed numbers into changed equipment.

Common problems

Problem	Action
Search cannot start	Read the input or coupling blocker and correct the named field.
Ensemble years are missing	Fetch the required window in Weather; a single imported year does not provide that cache.
All candidates fail one limit	Check the input evidence, equipment choices and search bounds before changing the limit.
A proposal is stale	Run the sizing study against the current model again.
A selected result differs from the editor	Check design version and run identity; a search does not rewrite every open draft.

Use Sensitivity to test whether a choice remains useful when assumptions change. These controls describe the September 2026 development build.